{"id":"arxiv:2609.30037","kind":"preprint","title":"AERIAL: Adversarial Evaluation of Robustness in Accuracy-Preserving Low-Precision EEG Decoders","authors":["Saim Rehman","Muhammad Shafique"],"author_count":2,"year":2026,"published":"2026-09-24","venue":null,"cited_by":0,"doi":null,"arxiv_id":"2609.30037","nct_id":null,"source_url":"https://arxiv.org/abs/2609.30037","archive_url":"https://space-kitty.com/papers/arxiv:2609.30037","updated":"2026-09-24","pmid":null,"pmcid":null,"openalex_id":"W7214341094","pdf_url":"https://arxiv.org/pdf/2609.30037","abstract":"Deployment-oriented compression is attractive for resource-constrained brain--computer interfaces (BCIs), but whether it changes adversarial vulnerability remains unclear. On BCI Competition IV-2a, we compare 32-bit floating-point (FP32) EEGNet and ShallowConvNet models with global magnitude pruning and simulated INT8 post training quantization (PTQ) and quantization-aware training (QAT) across nine subjects and three seeds. Simulation provides differentiable quantize--dequantize models for white-box attacks and gradient analysis, while native TensorRT deployment is used for validation. Accuracy-preserving compression does not improve direct robustness: at $ε=0.005$, EEGNet PGD accuracy remains 22--24\\% across FP32, 50\\% pruning (P50), PTQ, and QAT. However, P50 reduces bidirectional transfer efficiency to 0.963/0.928 (FP32$\\rightarrow$P50/P50$\\rightarrow$FP32), versus 0.994/0.997 for PTQ; the same trend holds for ShallowConvNet. Gradient alignment shows a corresponding separation, while native PTQ agrees with simulated clean/adversarial predictions in 95--98\\% of cases. These results show that direct robustness, adversarial transfer, and deployment efficiency are distinct properties of compressed EEG decoders.","abstract_license":"CC0-1.0 (arXiv descriptive metadata)","license":null,"is_open_access":false,"retracted":false,"peer_reviewed":false,"categories":["cs.CV","cs.CR","cs.LG","eess.SP"],"details":null,"provenance":"arxiv","attribution":"Abstract: arXiv descriptive metadata (CC0). Thank you to arXiv for use of its open access interoperability. This service was not reviewed or approved by, nor does it necessarily express or reflect the policies or opinions of, arXiv.","topics":[{"slug":"brain-computer-interfaces","name":"Brain-Computer Interfaces","domain_slug":"neuroscience"}],"cited_in_dossiers":[]}